BenchCI Documentation

BenchCI is a hardware validation tool for embedded systems. It lets you define a bench in bench.yaml, describe tests in suite.yaml, and run those tests locally or through a remote BenchCI Agent.

What BenchCI can do

BenchCI can:

• flash firmware automatically

• validate device behavior over UART, Modbus RTU, Modbus TCP, and CAN

• control or observe GPIO through local_gpio, remote_gpio, or mock_gpio

• run repeatable real-hardware tests in CI

• produce structured artifacts per run

Core model

BenchCI is built around:

• bench: the full execution environment

• nodes: named participants such as dut, controller, or gateway

• transports: UART, Modbus RTU, Modbus TCP, or CAN

• GPIO: logical input/output lines defined per node

• suite steps: declarative actions such as flash, reset, send_uart, expect_uart, gpio_set, and expect_can

Execution modes

BenchCI supports:

• local execution on the machine connected to the hardware

• remote execution through a BenchCI Agent

• uploaded-bench remote runs

• registered-bench remote runs using a persistent bench_id

• split deployments where GPIO is controlled through a remote Linux machine

Typical flow

bench.yaml + suite.yaml
        ↓
   benchci run
        ↓
 local runner or Agent
        ↓
 real hardware bench
        ↓
 logs + results.json

Documentation

• Installation
• Quickstart
• BenchCI CLI
• Bench Configuration
• Suite Configuration
• BenchCI Agent
• BenchCI Architecture
• GitLab CI Integration
• Linux GPIO in BenchCI
• BenchCI Examples
• 📦 Example Scenarios
• 🧠 How to Use These Examples
• ⚠️ Important Notes
• 🚀 Recommended Learning Path
• 🎯 Summary

Where to start

For first-time setup, read:

1. installation.md

2. quickstart.md

3. cli.md

If you want to run remote hardware infrastructure, continue with:

4. agent.md

5. gitlab_ci.md